A Retinex based GAN Pipeline to Utilize Paired and Unpaired Datasets for Enhancing Low Light Images

Weligampola, Harshana; Jayatilaka, Gihan; Sritharan, Suren; Godaliyadda, G. M. R. I.; Ekanayaka, Parakrama; Ragel, Roshan; Herath, Vijitha

doi:10.1109/mercon50084.2020.9185373

Cited by 9 publications

(3 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, the refinement network is also used for noise removal. Weligampola et al [ 38 ] proposed a novel deep learning pipeline, in which CNNs and GANs are optimized to minimize standard losses and adversarial loss. The proposed model divides the enhancement process into two parts; the decomp-net decomposes the images into reflectance and illumination based on Retinex, and it pays more attention to local information.…”

Section: Related Workmentioning

confidence: 99%

End-to-End Retinex-Based Illumination Attention Low-Light Enhancement Network for Autonomous Driving at Night

Zhao

Yong

Jian

2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Low-light image enhancement is a preprocessing work for many recognition and tracking tasks for autonomous driving at night. It needs to handle various factors simultaneously including uneven lighting, low contrast, and artifacts. We propose a novel end-to-end Retinex-based illumination attention low-light enhancement network. Specifically, our proposed method adopts multibranch architecture to extract rich features for different depth levels. Meanwhile, we consider the features from different scales in built-in illumination attention module. We encode reflectance features and illumination features into latent space based on Retinex in each submodule, which could cater for highly ill-posed image decomposition tasks. It aims to enhance the desired illumination features under different receptive fields. Subsequently, we propose a memory gate mechanism to learn adaptively long-term and short-term memory. Their weight could control how many high-level and low-level features should be reserved. This method could improve the image quality from both different feature scales and feature levels. Comprehensive experiments on BDD10K and cityscapes datasets demonstrate that our proposed method outperforms various types of methods in terms of visual quality and quantitative metrics. We also show that our proposed method has certain antinoise capability and generalizes well without fine-tuning when dealing with unseen images. Meanwhile, our restoration performance is comparable to that of advanced computationally intensive models.1

show abstract

Section: Related Workmentioning

confidence: 99%

End-to-End Retinex-Based Illumination Attention Low-Light Enhancement Network for Autonomous Driving at Night

Zhao

Yong

Jian

2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

show abstract

“…Image decomposition based on this model attempts to generate reflectance and illumination maps. This has been proven useful in applications such as lighting enhancement [3,4]. The major drawback of retinex models is their agnosticism to object surface geometries and the complicated physical phenomena related to light reflection.…”

Section: Introductionmentioning

confidence: 99%

An Optical Physics Inspired CNN Approach for Intrinsic Image Decomposition

Weligampola

Jayatilaka

Sritharan

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

Self Cite

View full text Add to dashboard Cite

Intrinsic ImageDecomposition is an open problem of generating the constituents of an image. Generating reflectance and shading from a single image is a challenging task specifically when there is no ground truth. There is a lack of unsupervised learning approaches for decomposing an image into reflectance and shading using a single image. We propose a neural network architecture capable of this decomposition using physics-based parameters derived from the image. Through experimental results, we show that (a) the proposed methodology outperforms the existing deep learningbased IID techniques and (b) the derived parameters improve the efficacy significantly. We conclude with a closer analysis of the results (numerical and example images) showing several avenues for improvement. Index Termsintrinsic image decomposition (IID), convolutional neural networks (CNN), Phong model.

show abstract

“…Image decomposition based on this model attempts to generate reflectance and illumination maps. This has been proven useful in applications such as lighting enhancement [3,4]. The major drawback of retinex models is their agnosticity to object surface geometries and the complicated physical phenomena related to light reflection.…”

Section: Introductionmentioning

confidence: 99%

An Optical physics inspired CNN approach for intrinsic image decomposition

Weligampola,

Jayatilaka,

Sritharan

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

A Retinex based GAN Pipeline to Utilize Paired and Unpaired Datasets for Enhancing Low Light Images

Cited by 9 publications

References 23 publications

End-to-End Retinex-Based Illumination Attention Low-Light Enhancement Network for Autonomous Driving at Night

End-to-End Retinex-Based Illumination Attention Low-Light Enhancement Network for Autonomous Driving at Night

An Optical Physics Inspired CNN Approach for Intrinsic Image Decomposition

An Optical physics inspired CNN approach for intrinsic image decomposition

Contact Info

Product

Resources

About